As noted at the outset, the Bayesian approach provides a coherent framework for permitting a decision maker’s judgment, expressed as prior beliefs, to enter the cost-of-equity estimation. A key feature of those prior beliefs explored in this study is the degree of mispricing uncertainty (cra). We set the prior mean of the pricing error (<5) equal to zero, but that specification could be relaxed, as discussed previously (Section I).
In particular, the prior mean for the pricing error could depend on one or more characteristics of the firm. The posterior mean for a is then adjusted away from that non-zero prior mean and toward a. and the degree of that adjustment would depend on the prior parameters crQ and E(cr2) in essentially the same manner as discussed previously.
Given the imprecision associated with estimates of factor premiums, found here and in previous studies, it seems essential that those quantities be estimated using as much information as possible. Our methodology allows that information to include series whose histories are longer than those of the factors—over twice as long in this study. Wre find that the additional information in those series produces posterior means for the factors, and thus for that differ substantially from those based on the factor histories alone.
We also find that, even after incorporating the additional information in series beginning in 1926, uncertainty about factor premiums still makes the largest contribution to overall uncertainty about the expected excess return (in the absence of uncertainty about a), although uncertainty about betas is nearly as important for the typical individual stock. The priors for the factor premiums are specified in this study as diffuse (non-informative).
One might instead be able to construct reasonable informative prior beliefs about one or more of the factor premiums, and the posterior uncertainty about a stock’s expected excess return would then no doubt be less than we report. Alternatively, introducing additional historical data, possibly within a different stochastic setting, might also prove helpful. In general, the uncertainty about factor premiums present in cost-of-equity estimation offers payoffs to future research. Electronic Payday Loans Online